|
From: <kot...@cs...> - 2007-06-28 19:41:08
|
Hi, 1. I have one question concerning cachegrind simulation on multicore. I execute a multithreaded programme on a 8 core Opteron board. I notice on the output from the top command that all 8 cores are utilised upon the creation of more than 8 threads. On the contrary, when I profile it with valgrind the cores utilized ,with the same configuration as prevrious, drop from 8 to 2-4. My first question is why I have this results? 2. Furthermore, is there a way to simulate all the caches on the multicore machine and gather results per cache? If not, what is the most appropriate way to do experiments into that direction? Thank you in advance Regards Christos |